Smoke and Mirrors

In drawn endings, one could offer a draw. Or, as Levon Aronian showed me in 2013, playing 36 moves in a dead-drawn ending, one can offer your opponent opportunities to err. Such was my guidance in a rapid game this morning.

After 47 moves, I had a slight advantage, but missed my best chance.

White to move

48.Rh6! would have maintained an advantage. For instance, 48...Raxh7 49.Rb5+ Kc7 50.Rxh7+ Rxh7 51.Rxa5+ and the three to one majority with one rook each offers White good prospects of victory.

I played 48.c3 and my advantage slipped away.

Ten moves later, I was certain the game was headed for a draw.

White to move

I played 58.a5+. Later, after my 71.b6, we have reached a technical draw that one finds in all the endgame books.

Black to move

Black played the only move, which is not hard to find, 71...Rh8. I proceeded to shuffle my king about and we reached this position after 79.Ka6.

Black to move

Clocks were running and the game continued with a blunder on each side.

79...Rg6?? (79.Rg8 was required) 80.Rh7?? (instead of 80.Rf8#).

If my opponent erred once, he might do so again.

Black to move

After 90.Rd7

90...Kc8?! 

This move is fine, although it suggests that Black may be experiencing some confusion.

91.Rd4 Re6??

My chance!

White to move

92.Ka7!

There are other winning moves as well.

92...Re7+ 93.Ka8 Re8 94.b7+ Kc7+ 95.Ka7 and Black resigned.

Must've been a Mouseslip

Online play this morning was odd. First I could get no advantage against a 900 rated player and that game was aborted because the Arena time ran out. Then, I was paired against a 1900+ (top .05% on a site that is 90% beginners) whose play left me confused.

Beginning at move five, my opponent's play was mysterious. We had reached a fairly normal position that can arise when White opts for the Catalan.

Black to move

I have had this position at least 283 times previously in online play, usually, but not always with White. In these games, both White and Black have performed within three points of their average rating with 138 White wins, 116 Black wins, and 29 draws. I have played 5...cxd4 and 5...Nc6, the two most popular moves.

5...g6?!

Only on Lichess, where the games database is huge, do I find any prior games with this move.

6.cxd5 exd5 7.O-O Bg7 8.dxc5

Black to move

8...Bf8? 9.b4

9. Nc3 was better. The one Lichess game with 8...Bf8 continued 9.Be3. It was a bullet game.

9...a5 10.Bd2?

10.Qa4+ Bd7 11.b5 Bxc5 and White is much better.

10...axb4 11.Bxb4 Nc6 12.Bc3

Black to move

12...Be7

White's poor play has restored Black to equality, but now White again has an advantage.

12...Bxc5 was the obvious move.

13.Nd4?!

13.Qc1 sets up a tactic to defend the pawn. To wit, 13...Bxc5 14.Bxf6 Qxf6 15.Nc3 Be6 16.Nxd5+

13...Bxc5 14.Nxc6

And after 14...bxc6, the game is interesting and roughly equal.

14...Qd6??

White to move

15.Nd4 Bxd4 16.Bxd4 Ra4 17.Qxa4+ and Black resigned.

I had thought that 14...Qd6 was a mouse slip, or maybe a premove, but the 5.5 seconds Black used rules out premove. It was a strange move in a strange game. 16...Ra4, then, looks like Black trying to lose.

My next game was against a player in the high 1600s and featured some strange maneuvers also, but not such that dramatically altered the evaluation. Here, though, I thought I had trapped White's queen.

Black to move

24...Rf6

24...Rxf5, which I considered briefly is the engine's choice.

25.Nxd6?

White should have played 25.Qxd6 Rxd6 26.Bxf4 and White will have a rook, bishop, and pawn for the queen.

I went on to win this game, too, although my opponent proved resourceful without a queen.

Far more satisfying than this morning's games was one that I played yesterday. At move 11, I had the opportunity to win a pawn, but spend a few seconds making sure that I was not missing a zwischenzug. 

Black to move

11...Ncxd4 12.Nbxd4

12.Nfxd4 Nxd4 13.Bd3 might be better.

12...Nxd4 13.Nxd4 Qxd4 14.Bxb4 Qxb4 and I nursed the extra pawn into the endgame, eventually winning.

Place a Rook

In the following diagram, where can White place a rook in order to have a drawn position with Black to move?

I found a game with this position and a White rook on h8 using the report feature of ChessBase 17. I had been playing with a position derived from Ending 57 in 100 Endgames You Must Know by Jesus de la Villa. Aside from the position that he presents, I wanted to understand where else the rook might be placed and still draw.

De la Villa's Ending 57 has this position with White to move.

After 1.Rd8, Black has two moves that hold a draw.

Before looking at de la Villa's solution, I placed Black's rook on a1 and made White's move Rd8. Then, I played Black against Stockfish on my iPad, trying a couple of variations. Then, I checked my ideas with ChessBase.

This was one of the resulting positions.

Black to move

4...Ke6 is the only drawing move.

This was another position in my play against the engine.

Black to move

Here, the side check is Black's only chance. 2...Ra7+

Using the report > similar endgames feature of ChessBase turned up far more games than I have time to look through. I favored those that were decisive, certain that I could find examples of games where a player who could have drawn instead lost.

No doubt both players were in considerable time pressure when Black erred from this winning position.

Black to move

Black played 110...Rf1?? 

110...d2 wins. To wit, 111.Re8+ Kf4 and White soon runs out of checks.

Nonetheless, Black won the game because his young opponent missed the drawing idea.

In the diagram at the top of this post, where must White place the rook?

Shifting Focus

For 34 consecutive days, I spent the first part of my day solving puzzles on chessdotcom. At some point during this time, it became a daily commitment. That commitment ended yesterday when I solved a single exercise from the book Improve Your Chess Tactics (2012) by Yakov Neishtadt. This book has sat idly on my bookcase for a year or more, but recently I started reading it. A few days ago, I solved the first five exercises after my chessdotcom session. Yesterday morning, I opted to start with the book.

Black to move

The position arose in Panno -- Bravo Sedamanos, Fortaleza (1975). It took me a few minutes, but I found the correct moves. The idea was known before solving because Nieshtadt organizes the book by themes. Hence, solving primarily involved calculation to verify the implementation of the deflection idea and the correct sequence. Most of my effort was expended finding the best reply by White.

After writing my answer, I checked my answers to the first six exercises in the back of the book. I failed one.

White to move

The position is presented by Nieshtadt as from Belov -- Ongemakh, Narva 1984. After seeing that I failed, I instantly saw the refutation of the move I had written.

Solving from books differs from solving for rating with online puzzles. Both methods have their benefits. I remain committed to doing both, but a daily commitment to one gets in the way of the other.

During my 34 day run from 18 May to 20 June, I attempted 599 puzzles, correctly solving 356. My rating started at 2993 and ended at 3051. It peaked at 3143 on 15 June and was as low as 2849 on 30 May.

I will return to the website and make a push to get the rating back above 3100 now that I've crossed that mark a couple of times. Also, I like to show my students that I solved more puzzles in the past year than the number of games I've played, although that number is considerable.

My focus has shifted towards books. Neishtadt, Improve Your Chess Tactics is one of the better tactics books available. Someone recommended it to me in a comment on this blog several years ago. It was a good recommendation for which I am grateful.

One Month of Puzzles

My training has rarely been consistent. Much of the time, it is not even training. I enjoy playing chess and enjoy studying it, too. Sometimes I set goals. Most of the time I achieve these goals.

Ten years ago, I began to abandon one of the few that I failed to achieve: a USCF rating above 2000. I had peaked at 1982 in 2012. In 2013, I recorded a series of training logs here on Chess Skills. In February 2014 and again in August, I fell below 1900. I rose back above that mark once more in August 2015 when I won a weekend Swiss for the first time (see “Winning an Open”). My most recent first place finish in a weekend Swiss was in 2023 (see “Misevaluation”). In that event, I was playing for the enjoyment of the game. Over the previous months, regular puzzles had been a habit and contributed to my enjoyment of the game during the tournament.

My USCF rating graph

Although my ambitions to cross 2000 USCF have dissipated, improving my knowledge and skills has continued. The goals are small and training related. I seek small achievements, such as getting my online rapid rating above 1800, playing through every game in a single issue of Chess Informant, memorizing a batch of classic games, or reaching the next century mark in puzzles.

I enjoy learning. The process of gaining knowledge or microskills* is often its own reward.

Nearly a month ago, I publicized a rating goal with chessdotcom puzzles. Dean Arond questioned the benefits: “but does it translate to your USCF rating?” The next four days saw my puzzles rating fall 200 points, but then some consistency brought it back up and I crossed 3100 on 9 June. Posting a link to Chess Skills, where I marked the achievement, provoked more skepticism and an informative discussion with Jon Jacobs. He offered a link to a 2008 blog post where his critical comment led to considerable discussion concerning the merits of following the tactics training regimen advocated by Michael de la Maza. There is a lot to process at that link and I appreciate the perspective that Jacobs offers.

The past month, since 18 May, I have attempted a minimum of ten rated puzzles every morning. During the solving, I have addressed a specific weakness that affects my playing performance, as well as my habits while solving tactics. I often see an idea and play it instantly or after superficial calculation. During my puzzle solving the past month, I have focused on accuracy. If I suspect a checkmate is present, I work it out to the end, laboring to find all manner of resistance.

Progress has been up and down. My tactics rating today (3027) is lower than it was on 18 May (3038). Nonetheless, I reached new highest ever ratings half a dozen times or more with a current peak four days ago at 3143. More important than rating level is the increase in my percentage of success. My puzzle accuracy (percentage solved correctly) over the life of the site (more than 15,000 puzzles) increased from 53% to 54%. In the past month, I have exceeded 70% accuracy during 16 sessions.

There have been three sessions where accuracy was below 50% and these were the longest sessions. Often I enjoy the process of solving puzzles, but experienced frustration on those days. Plans to solve daily seemed more of a burden on those days.

More than likely, I will not continue a daily regimen ten or more puzzles on chessdotcom much longer. There are other avenues for tactics, and other forms of productive learning. For instance, a book, 100 Soviet Chess Miniatures (1963) by P. H. Clarke, has been on the side table in my living room since January. I've gone through the first 23 games.

*I use this term for specific skills that are part of my teaching curriculum, such as specific pawn endings that I want my students to master, the Lucena and Philidor rook endings, or the not yet achieved queen vs. rook ending I plan to work on more in the near future. I expect to play against the computer every rook endgame in Jesus de la Villa, 100 Endgames You Must Know, new edition (2015). This practical book study is an example of microskill development.

3100+ Puzzle Rating!

This morning, I reached a new peak puzzle rating of 3121 on chess.com. Nearly 18 months ago, I crossed 3000 for the first time (see "Working the Puzzles"). On May 24 and 25 this year, I hit new peaks of 3071 and 3079. Then in answer to Max Illingworth's question on Facebook, I publicized a goal of reaching 3100. Illingworth posted in his group, Adult Chess Improvers, "New week, new start. What's on your mind today?" I replied, "Hoping to get to 3100 chessdotcom puzzle rating. Hit a new peak of 3079 this morning."

After publicizing this goal, my rating fell more than 200 points in two sessions over the next four days. That's smaller than my 428 point drop from 2975 to 2547 last August, but it is still significant enough that I began to regret posting the goal. Still, I knew what to do: keep solving ten or more puzzles each day, being careful to calculate accurately. Persistence leads to success. Since 18 May, I have started each day with puzzles and mostly solved with a great deal of care.

This morning's session started with four correct of the first five and a rating of 3093. Then, this this puzzle.

White to move

I saw immediately a sequence that failed, then I reversed the first two moves. The process of calculation and then moving took me 33 seconds. My rating jumped 15 points to 3108. I missed the next puzzle, but then got two correct and stood at 3121. It might have been time to quit if not for my resolve to solve at least ten each day. I could not quit after nine.

The next puzzle dropped me 13 points, so I kept going, failing three more. Now at 3073, one point below the day's starting rating, I felt that I had to continue. I was able to quit after solving the next five.

A Mednis Error

Edmar Mednis, Practical Rook Endings (1982) is proving its value as I am using positions from it with a student who wisely requested a focus on rook endgames. Last week, I spent one early morning studying Mednis’s analysis of Ljubojevic — Gligoric, 1979. Ljubomir Ljubojevic won game nine of their match when Svetozar Gligoric missed a critical defensive resource. However, this resource was possible because of an inaccurate move by Ljubojevic.

After finding the analysis instructive, I set up the game to play against Stockfish. The engine showed me that had Ljubojevic played the recommended move, there was another defensive resource available to Gligoric. This resource refutes Mednis’s claim that Ljubojevic missed a win when he pushed his g-pawn before playing 68. c4!

Mednis’s analysis begins after 66…Rg4.

White to move

The game concluded 67.g7 Kb7 68.c4 Rg2 69.Kc3 and Gligoric resigned.

According to Mednis, White should have played 67.c4, when White wins after 67…Rg3+ 68.Kc2 Kb7 69.g7 Ka7 (there is a typo in the book giving Kc7, which the prose makes clear is not the intended line, as it loses immediately to 70.Ra8) 70.Kd2.

Black to move

From Mednis's Analysis

The idea is that White’s king will move towards the rook, up the f-file, and then over to capture Black’s c-pawn. All of this is explained clearly and accurately by Mednis. White’s c-pawn then destroys Black’s defensive fortress with the king safe from skewers by remaining on a7 and b7.

However, Stockfish, which relies on deeper calculation rather than endgame principles, showed that Black can play 68…Kd7!! Playing this position out leads to a drawn pawn ending.

My game against Stockfish from the diagram at the top of this post went thus:

67.c4 Rg3+ 68.Kc2 Kd7! 69.g7 Ke7 (69...Ke6 is also drawing)

White to move

70.Ra8 

I might have tried 70.Kd2 Kf7 71.Rf8+ Kxg7 72.Rf5, but exploring this line later showed that all roads lead to a draw. Still, Black must either understand the needs of the position or calculate with the sort of depth employed by Stockfish.

70...Rxg7 71.Ra7+ Kf6 72.Rxg7 Kxg7 73.Kb3

Black to move

73...Kf6 74.Ka4 Ke5 75.Ka5 Kd6 76.Kb5

Black to move

All of my students learn early on how to draw this position with Black.

76...Kd7 77.Kxc5 Kc7=

Having played against the computer many of the positions from Mednis, Practical Rook Endings, I have observed that his analysis usually holds up to computer checking. In this particular ending, however, he did not look deep enough into a possible pawn ending had Gligoric opted to ignore the usually prudent movement of the king to the crucial a7 and b7 squares.

While ruminating over this analysis during my morning walk with my dog, it occurred to me that analysis presented by Mednis may not have originated with him. A quick check of the endgames published in Chess Informant confirmed my hunch. All of Mednis's analysis appeared in the endgame section of Informant 27. There it is credited to two of the founding members of CI's editorial team: Borislav Milic and Aleksandar Bozic.

Distraction

This morning’s tactics session was suboptimal. After solving the first one correctly, I did so poorly that I dropped my rating 98 points from 3038 to 2940. I continued to solve with increased determination, ending at 3004. The last exercise, which I solved correctly in 4:59, offers a case study in distraction. How often do we focus on one area of the board, completely oblivious to what else is happening somewhere else.

I spend most of five minutes trying to make something work on the queenside, noting the threatened knight fork if my queen strayed. I looked at the contact between the bishops. In the last few seconds after expending so much time looking at unproductive lines, I noticed Black’s fatal problem on the kingside and executed the simple mate in two.

Self-propelled Pawns

For several years, I have used the term "self-propelled pawns" for two connected passed pawns in rook endings. I do not know whether I made up the term or read it in a book. Nonetheless, it is usually fairly easy to win such positions. Reuben Fine, Basic Chess Endings (1941) wrote, "connected pawns always win unless they are blockaded" (317).

During the 25% off sale earlier this week, I took the plunge and upgraded from ChessBase 15 to ChessBase 17. The new version lacks a feature that I regularly use: endgame keys. However, the help files suggest a different method of finding particular types on endings: similar endgames under report (see image).

I tried the new method this morning and was not surprised that Megabase 2024 contains many thousands of games with two connected passed pawns in rook endings. Naturally, I started looking for games when the stronger side failed to win. In most cases, the ending was misplayed.

I had the position above a year ago in a 16 minute game on Playchess. I finished the game checkmating my opponent 20 moves later using 3 seconds per move or less on all but five moves. 9 seconds was the longest.

I do not have access to the time per move nor time remaining for those games that were drawn, although winnable. Even so, perhaps the errors are instructive.

Black to move

This position was reached in a game played in 1998. Black, rated 1355, played 69...f4 and the game was drawn after 70.Rf8. 69...Kg4 would have been my choice and Stockfish confirms that Black is winning.

In a game played in 2003 between two 1500s, the two pawns plus rooks ending was reached on move 47. Black made progress advancing his pawns over many moves. Even here, after 87.Rd5+, Black was winning.

Black to move

87...Rd3 88.Rxf5 Rf3+

This move throws away the win. 88...e3 was the only way to keep the win in hand.

Black to move

Black, rated 1598, spoiled the win with a move that reminded me of a position from Kling and Horwitz (1851) and posted as "Elementary Technique" a month ago.

62...Re5+

I would have played 62...Re1, which is one of several winning moves.

63.Rxe5 fxe5 64.Kxg5 and drawn.

White could have erred with 64.Kxe5, when the idea in Kling and Horwitz would secure the win for Black.

The final example is from a 2010 game between two stronger players in the championship of Chili.

Black to move

55...Kg8

There is nothing wrong with the move in the sense that it does not spoil the win. However, 55...h4 can be played. However, 56.Ra7 forces Kg8. 57.Rg7+ Kf8 58.Rh7 Ke8 and this line begins to resemble the game as played.

56. Rg7+ Kf8 57.Rh7 Ke8 58.Ke6 Kd8 59.Kd6 Kc8 60.Kc6 Rc4+ 61.Kd5

Black to move

61...Rc7 spoiled the win, which is not entirely simple. Tablebases assure us that 61...Rf4 leads to checkmate in 42 moves. Black's king is still cut off, though.

White found 62.Rh6! and a draw was agreed.

Exchanging rooks leads to an elementary winning pawn ending, while 62.Rh8 allows the Black king to become active.

Easy Exit

This morning in an online 10 minute game, I knew I had an advantage, but suffered a failure of confidence without a clear checkmate. I forced a draw. The rating difference was such that the draw lifted me 6 points.

I had Black

White to move

32.a6

While replaying from the diagram against Stockfish, the machine played 32.Qc6. A queen ending was reached after the subsequent moves: 32...Ng3+ 33.Kh2 Ne2 34.Rxe2 Qxe2 35.Bc1 Qd1 36.Qc5 Qxb3 37.Bd2 h5 38.Bxb4 Rb8 39.a6 Rxb4 40.a7 Ra4 41.Qc8 Kh7 42.a8Q Rxa8 43.Qxa8

Black to move

Analysis diagram

My failure to convert the advantage in this ending highlights an area that deserves some study.

Returning to this morning's game.

32...Qe1+

32...Ng3+ is better.

33.Kh2 Qg3+ 34.Kg1

Black to move

At this point, I knew that I had an advantage and spent 58 seconds looking for a clear path forward. 

34...Qe1+

I opted to force a draw.

34...Ng5 created threats, but did not force matters towards a favorable conclusion.
I missed the option of returning to a previous position and easily winning White's rook. To wit, 34...Qe3+ 35.Kh1 Ng3+ 36.Kh2 Ne2 and White must give up the rook to stop checkmate.

The same situation would have occurred after 32...Ng3+

After the rook exchanges itself for the knight, Black's queen forks bishop and pawn.

White to move

I had no difficulty winning this position against Stockfish in my second battle with the silicon beat today.

Inspired by Ruy Lopez

This morning I had a familiar position although the opening moves were not anything I recall playing before. The position was familiar because it had the same critical elements (pattern) found in a game Ruy Lopez played in 1560 and that appears in my book, Checkmates and Tactics (2019). The exercises in this book are those I've been using with scholastic chess players since 2006 as part of my award curriculum.

I had White in the 10 minute game.

1.e4 e6 2.c4 c5 3.Nf3 d5 4.exd5 exd5 5.d4 dxc4?! 6.Bxc4 Bg4?

Bad pins often backfire.

White to move

7.Bxf7+ Kxf7 8.Ne5+ creates a discovery against the bishop as well as forking king and bishop.

8...Ke8 9.Qxg4 Nf6 10.Qe6+

The game is continuing along the path shown by Lopez.

10...Qe7

White to move

11.Qc8+ Qd8 12.Qxd8

I considered 12.Qxb7, which Stockfish prefers, but reasoned that Lopez's combination was good enough to give me a clear advantage.

12...Kxd8 13.Nf7+

What we have of Lopez's game ends with this fork.

13...Ke8 14.Nxh8

My opponent made me play all the way to checkmate. Along the way, I won the other rook through another combination.

White to move

24.Rxe6

24.Rxb7+ is simpler and better.

24...Kxe6 25.Nc7+ Kf5 26.Nxa8

It's not often that I get to fork both rooks with my knights in the same game.

Lopez's game started with a King's Gambit Declined. His opponent is given in ChessBase Mega as Giovanni Leonardo da Cutri (spelling and name sequence differs from source to source).

1.e4 e5 2.f4 d6 3.Bc4 c6 4.Nf3 Bg4

The bad pin.

White to move

This position is number 21 in Checkmates and Tactics, available through Amazon both in print and Kindle versions. It is the third exercise in my Bishop Award set.

6.Bxf7+ Kxf7 7.Nxe5+ Ke8 8.Qxg4 Nf6 9.Qe6 Qe7 10.Qc8+ Qd8 11.Qxd8

Lopez's move is best, which differs from my game.

11...Kxd8 12.Nf7+

Active King

In a ten minute game this morning, I struggled most of the game for equality. When my opponent erred in opting for a rook ending, rather than knight vs. bishop when his passed pawn likely would have cost me my bishop, I knew my drawing chances were good.

Then, while reviewing the game, I was surprised to see that a move that assured the draw was deemed an error by the website's game analysis feature.

Let's look at the critical position with normal diagrams (White on bottom).

White to move

I played 47...Kxf4, reaching an easily drawn ending that neither side should have difficulty playing.

This position, however, is winning for Black.

47...Rb2+ 48.Kf1 (Kg1 likely transposes) 48...Kf3

Leaving the f-pawn alone while creating mate threats was completely off my radar, perhaps because I knew there was no forced checkmate. The threats, however, gain something.

49.Kg1 Rb8+ 50.Kh2 Kf2

White to move

White is forced to move the rook and cannot attack my g-pawn.

51.Rh8 Rxb7 52.Rh6 Re7 

Black keeps mate threats alive.

53.Kh1

Black to move

Two of the roads that lead to victory from here are:

a) 53...Re6 54.Rh2+ Kg3 55.Rg2+ Kxf4 finally capturing the f-pawn!
b) 53...g5! 54.Rh2+ (54.fxg5 Kg3 and White must give up the rook to avoid checkmate) 54…Kf3

White to move

Black's g-pawn is untouchable due to the mate threat.

55.Rb2 Re1+ 56.Kh2 g4

I would have known that I was winning had I foreseen that I could reach this point.

It was good to draw a 2000+ rated player, but there was an instructive missed opportunity that I hope to remember.